AMD CPU speculation... and expert conjecture

Page 406 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

szatkus

Honorable
Jul 9, 2013
382
0
10,780

logainofhades

Titan
Moderator
To not recommend an AMD CPU at any price point is a foolish statement to make. If on a budget, AMD is the better option so you can get a better GPU, which matters far more than CPU in 99% of games out there. I would take an FX 6300 and bump it to an FX 6350 over an i3 anyday.

I would take an FX 8320 over a locked i5 4670. It isn't worth the $75 price premium. That $75 would be better spent on a faster GPU. i7's are even more overpriced. 4770k isn't worth twice the price of an FX 8320 and even the locked i7's are stupidly priced when you can get a 1230 v3 for about $50 cheaper than a 4770.
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


[Edit: Nvm didn't see the specs were hidden/collapsed]
 

Rum

Honorable
Oct 16, 2013
54
0
10,630


One cannot deny the results "IF they are true'! For all we know it can be another intel puff piece
But why so smug about it?.. If anything we should all be disappointed that AMD is not competing better than they are, all people like you and Hafijur do is encourage people to keep spending their money on the evil blue giant to continue squashing AMD.
 
that hardcoreware article says that the core i3 4340 costs $160....just.. no. core i3 over $99-120 aren't worth buying since lowest core i5s cost only 20-30 bucks more. cpus like core i5 4430, 4440, fx8320 are better buys than the i3 4340.
the igpu benches are done in 720p res. i don't think hd4600 can handle 1080p.
edit: the only way to make the core i3 4340 barely palatable is to combo it with an entry level h81 motherboard to keep total cost roughly under $220 (jetway to teh rescue) and pray that the mobo doesn't clip top 100-200mhz on load.
or use it with a cheap mini itx board and a higher end gaming card in a smaller form factor pc. fx can't do mini itx and apus don't have the cpu performance. only lower core i5 will be better (they'd be better choices regardless). may be until kaveri.
 


I'm smug because I've been the champion for IPC being far more important then the number of cores since the BD rumor thread, and have been run out of the forums on several occasions for my stance the highly clocked i3's, in games, can outperform the FX-8xxx series of chips from AMD, even in well threaded games. Now we have an i3 doing exactly that. So yeah, just a LITTLE bit smug right now.

I took a closer look at this GeekBench. In fact it isn't memory bounded.

http://browser.primatelabs.com/geekbench3/compare/28129...

So what's going on?

http://browser.primatelabs.com/geekbench3/compare/23765...

Identical CPU, OS, chipset, motherboard.
What about turbo? Frequency depends on chip's temperature. We only know that it could be somewhere between 3.4Ghz and 4.0GHz. Even temperature in room can change results slightly.

Simple:

For the first case:

Memory 8192 MB DDR3 SDRAM 534MHz 8192 MB DDR3 SDRAM 667MHz

Accounts for the memory results.of about 15% performance toward the faster RAM. Rest is just statistical noise for the most part.

for the second case:

BIOS American Megatrends Inc. F1 American Megatrends Inc. F4b
Memory 32768 MB DDR3 SDRAM 666MHz 8192 MB DDR3 SDRAM 1002MHz

Different BIOS could be playing a part for the non-memory based tests. Also possible is really shoddy resource management by the OS as RAM increases (which is QUITE possible actually).
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


[Edit: Nvm didn't see the specs were hidden]

It's like the opposite of TH review of the i7-4770K vs A10-5800K where the IGP of the i7 only won in Warcraft and lost in all the others. Even that used DDR3-1600 which handicaps AMD APUs.

What's the cause of the big change in performance? New drivers?
 

noob2222

Distinguished
Nov 19, 2007
2,722
0
20,860
metro.png


Makes one wonder how they achieved such massive differences in favor of intel cpus on hcw site. Xbit usually doesnt like to show this side of amd.

I call bs on the igp test.

http://us.hardware.info/reviews/4993/6/intel-core-i3-4330--i5-4440-review-affordable-haswells-benchmarks-igpu-crysis-3-1920x1080-medium
 

jdwii

Splendid




Ok i side with gamer half the time on this forum and honestly he is not pro Intel or Amd he is pro X86 performance, these screen shots if real do show an issue with Amd's current technology. In my opinion Bulldozer or the module design was a bad approach- Latency in a CPU is bad and when the cores share resources or fight for them you got a latency problem. Steamroller in simple terms made a module share less and share more efficiently as well as adding more resources which should help increase X86 performance and make this less of a problem with the design.
But where me and gamer disagree on is using all 8 cores for gaming which is going to happen on the consoles since they have to! They will find a way (mantle is just a step closer)
On current games Amd is behind even in BF4,Crysis 3 and so on but not to the point where the game is unplayable and not to the point where I can’t recommend the processor within a budget and actually besides AM3+ being a dead socket I would recommend a 8350FX plus a 212+ cooler over a I5 because of games using more cores.
I’m sorry juanrga but Amd only making a 4 core processor when the consoles have 8 and when they talked about more cores for quite some time makes no since

Edit-
Not to mention what you are saying for your reason is not right either to support a company to just keep them in business is just as wrong
 

Ags1

Honorable
Apr 26, 2012
255
0
10,790


Hmmmm, Intel IGP looking much stronger! Or maybe its all BS. I'd like to see richland and the haswell i3 run through my benchmark - I almost trust that.
 


The IGP test is actually easy to explain: 720p low versus 1080p medium. Also raises the specter of the more powerful CPU pushing a potentially weaker IGP along. Would be interested to see the same sites flip their resolutions/settings, and see if they start getting the same numbers...

The CPU numbers I *think* I can explain; easier to accept then a giant conspiracy. Some graphical settings are more CPU intensive then others, and depending which ones you flip on, you can bias performance significantly. Take our Metro example:

Xbit: 1920, SSAA [HOW MUCH???], High Quality
Xbit: 1280, No SSAA, High Quality
Hardcoreware: 1920, 2xSSAA, Medium Quality

High settings are obvious GPU bottleneck. So Xbit dropped the res, but kept high settings. This implies Tessellation is still enabled. My theory is that when you enable Tessellation, its too much total work for the i3 to process, and they get bottlenecked. Disable tessellation, and the i3s are no longer bottlenecked, and pull ahead. If I'm right, Tessellation basically kills i3 performance in Metro. Anyone know if anyone's benched this?
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


He mentioned the i3 GPU could overclock to 1.5Ghz, impressive. That would make quite a difference from the normal max 1150.

Also he's effectively overclocking the memory for the i3 (using 1866) which only specifies DDR3-1600. As we know memory plays a big role in APU performance. Richland supports 2133.

The performance gap definitely isn't as big as it used to be but if they're going to compare the flagship i3 to an A10 they should really use the flagship A10-6800K and the latest FM2+ platform.

There's no denying the Haswell core is strong. Who would pass up a fully unlocked i3?
 

Rum

Honorable
Oct 16, 2013
54
0
10,630


I am not advocating to support AMD just to strictly keep them in business, I think they do have some good products and with some financial support and backing they could have some great products. Either way AMD needs to quit screwing around and fix the problems that effect the cpu design and stop with the bandaid fixes, like higher clocks or blaming the software for everything.

You are right, I fail to see the logic as to why AMD would push the more cores mantra, then turn around and go back to quad cores? makes no sense at all!

As for the the "dead socket" I am curious why so many bring it up as such a horrible thing when intel creates a dead socket every new cpu release?
 
It makes perfect sense rum.

AMD is focusing on the mainstream market at this time, which is 90% of all computers sold. And the 4 core APU is not only the heart of that offering at this time, but also move forward AMD's goad of HSA which would bring the GPU power accessible to programers for general computing.

This is not a bad strategy at all. Im quite certain that once the launch bugs in Steamroller are ironed out, you will see server and enthusiast parts appear. Not sure if that will be AM3+ or FM2+, since there is nothing stopping them from launching an 8 core on FM2+ and consolidating the platforms. ]

If HSA succeeds, the APU will be remembered as a genius move. IF.....
 

Rum

Honorable
Oct 16, 2013
54
0
10,630


Oh I know why they are focusing on the mainstream market and it is a good plan! I was just referring to what some have said about the high end/ fx CPUs being dead and that it makes no sense why AMD would kill that off when it is starting to gain traction.

I can see that as a possibility and it would make sense to iron out the kinks before releasing the higher end stuff.

Lets hope the IF comes to pass and the tech is utilized to it's fullest!
 

noob2222

Distinguished
Nov 19, 2007
2,722
0
20,860
@gk

The bigger discrepancy is between the haswell i3 and ivy. Xbit is only a few fps where hcw is over 40% different.

Haswell isnt that much better than ivy in any other tests.

As for trying to defend the igp here is tomb raider at 900p low.

tombraider.png


Hcw has intel being 2x as fast as amd in that test.

Hcw numbers just dnnt look believable, especially somehow getting bf4 single player to look like that. http://www.techspot.com/review/734-battlefield-4-benchmarks/page6.html?mobile=false
 


They have never said that is what they are doing, but after getting kicked around after Barcelona and Bulldozers launch, that is what I would do. The mainstream versions of those chips were well received, in part because the buggy launch versions were better suited for that market.

The reason I believe the delay theory is no x86 server upgrade on the current roadmap either. I might be willing to buy AMD giving up on the high end desktop market, but the lucrative server market also? I doubt that. And the development of those two parts is closely related. The roadmap I saw also says piledriver on am3 and server through 2015. It doesn't exclude a die shrunk piledriver as a stopgap.
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


AMDs server market share has dropped to like 5%. They're under powered and suck more juice. Neither of which is great for that market place. Desktops can tolerate higher power CPUs but not so much when you're trying to link 1000s at a time.
 


All of which supports the idea they are going to work out the bugs in steamroller before they launch a half baked solution.

And even with a small share, its a growing market with high margins. It seems unlikely they would drop it so easily. Especially since the Bulldozer architecture was built with servers in mind. It doesn't even require huge investments since so much of the desktop tech is fairly easily ported.
 

jdwii

Splendid


Well they actually knew that when they released piledriver and bulldozer and phenom yet we still saw 6 cores and 8 cores.

Not really an excuse? Actually if we look at video cards we can probably ask why their still building high-end GPU's when their not meant for mainstream?
 

Master-flaw

Honorable
Dec 15, 2013
297
0
10,860

For the record, that quotes bug...wasn't me who said that.
 
True, jdwii. But the management of AMD and the market as a whole have changed in that time.

And need I point out that the "rush the server and enthusiast parts out first" strategy did not go well for either Barcelona or bulldozer. Perhaps a lesson learned?
 
Status
Not open for further replies.